Amazon is implementing a one-year moratorium on police use of its artificial intelligence software Rekognition amid a growing backlash over the tech company's ties to law enforcement. The company has recently stated its support for the Black Lives Matter movement, which advocates for police reform – using Twitter to call for an end to "the inequitable and brutal treatment of black people" in the US and has putting a "Black lives matter" banner at the top of its home page. But the company has been criticized as hypocritical because it sells its facial recognition software to police forces. Amazon has not said how many police forces use the technology, or how it is used, but marketing materials have promoted Rekognition being used in conjunction with police body cameras in real time. When it was first released, Amazon's Rekognition software was criticized by human rights groups as "a powerful surveillance system" that is available to "violate rights and target communities of color".

For an intelligent agent to flexibly and efficiently operate in complex environments, they must be able to reason at multiple levels of temporal, spatial, and conceptual abstraction. At the lower levels, the agent must interpret their proprioceptive inputs and control their muscles, and at the higher levels, the agent must select goals and plan how they will achieve those goals. It is clear that each of these types of reasoning is amenable to different types of representations, algorithms, and inputs. In this work, we introduce a simple, three-level hierarchical architecture that reflects these distinctions. The low-level controller operates on the continuous proprioceptive inputs, using model-free learning to acquire useful behaviors. These in turn induce a set of mid-level dynamics, which are learned by the mid-level controller and used for model-predictive control, to select a behavior to activate at each timestep. The high-level controller leverages a discrete, graph representation for goal selection and path planning to specify targets for the mid-level controller. We apply our method to a series of navigation tasks in the Mujoco Ant environment, consistently demonstrating significant improvements in sample-efficiency compared to prior model-free, model-based, and hierarchical RL methods. Finally, as an illustrative example of the advantages of our architecture, we apply our method to a complex maze environment that requires efficient exploration and long-horizon planning.

Online learning, or more specifically, the multi-armed bandit (MAB) problem, focuses on the task of learning the reward distributions from an unknown environment while simultaneously optimizing cumulative rewards over a fixed time horizon T. This problem has been studied extensively when the environment (i.e., reward distributions) is stationary over time, with numerous algorithms proposed to tackle the tradeoff between exploration and exploitation when making decisions (see Bubeck et al. 2012 for a comprehensive review). While the stationarity assumption about the reward distributions greatly simplifies the analysis, it does not hold in many decision problems in OR/MS and other fields when the environment is time-varying. For example, a fashion retailer should take into account the seasonal demand shift when setting the prices for apparels, and a hospital needs to consider the variation of the patient arrival rate when scheduling the medical staff. Despite the practical relevance, it is difficult to develop a learning policy for non-stationary rewards, especially when the dynamics can change arbitrarily over time. Recent studies (Besbes et al., 2015) have considered cases in which the environment does not change fast with respect to the length of the time horizon, e.g., when a budget sublinear in T is imposed on the total variation of the underlying reward distribution.

We introduce Bi-GNN for modeling biological link prediction tasks such as drug-drug interaction (DDI) and protein-protein interaction (PPI). Taking drug-drug interaction as an example, existing methods using machine learning either only utilize the link structure between drugs without using the graph representation of each drug molecule, or only leverage the individual drug compound structures without using graph structure for the higher-level DDI graph. The key idea of our method is to fundamentally view the data as a bi-level graph, where the highest level graph represents the interaction between biological entities (interaction graph), and each biological entity itself is further expanded to its intrinsic graph representation (representation graphs), where the graph is either flat like a drug compound or hierarchical like a protein with amino acid level graph, secondary structure, tertiary structure, etc. Our model not only allows the usage of information from both the high-level interaction graph and the low-level representation graphs, but also offers a baseline for future research opportunities to address the bi-level nature of the data.

Real-time video inference on compute-limited edge devices like mobile phones and drones is challenging due to the high computation cost of Deep Neural Network models. In this paper we propose Adaptive Model Streaming (AMS), a cloud-assisted approach to real-time video inference on edge devices. The key idea in AMS is to use online learning to continually adapt a lightweight model running on an edge device to boost its performance on the video scenes in real-time. The model is trained in a cloud server and is periodically sent to the edge device. We discuss the challenges of online learning for video and present a practical design that takes into account the edge device, cloud server, and network bandwidth resource limitations. On the task of video semantic segmentation, our experimental results show 5.1--17.0 percent mean Intersection-over-Union improvement compared to a pre-trained model on several real-world videos. Our prototype can perform video segmentation at 30 frames-per-second with 40 milliseconds camera-to-label latency on a Samsung Galaxy S10+ mobile phone, using less than 400Kbps uplink and downlink bandwidth on the device.

Drug repositioning is designed to discover new uses of known drugs, which is an important and efficient method of drug discovery. Researchers only use one certain type of Collaborative Filtering (CF) models for drug repositioning currently, like the neighborhood based approaches which are good at mining the local information contained in few strong drug-disease associations, or the latent factor based models which are effectively capture the global information shared by a majority of drug-disease associations. Few researchers have combined these two types of CF models to derive a hybrid model with the advantages of both of them. Besides, the cold start problem has always been a major challenge in the field of computational drug repositioning, which restricts the inference ability of relevant models. Inspired by the memory network, we propose the Hybrid Attentional Memory Network (HAMN) model, a deep architecture combines two classes of CF model in a nonlinear manner. Firstly, the memory unit and the attention mechanism are combined to generate the neighborhood contribution representation to capture the local structure of few strong drug-disease associations. Then a variant version of the autoencoder is used to extract the latent factor of drugs and diseases to capture the overall information shared by a majority of drug-disease associations. In that process, ancillary information of drugs and diseases can help to alleviate the cold start problem. Finally, in the prediction stage, the neighborhood contribution representation is combined with the drug latent factor and disease latent factor to produce the predicted value. Comprehensive experimental results on two real data sets show that our proposed HAMN model is superior to other comparison models according to the AUC, AUPR and HR indicators.

We present a new approach for sampling conditional measures that enables uncertainty quantification in supervised learning tasks. We construct a mapping that transforms a reference measure to the probability measure of the output conditioned on new inputs. The mapping is trained via a modification of generative adversarial networks (GANs), called monotone GANs, that imposes monotonicity constraints and a block triangular structure. We present theoretical results, in an idealized setting, that support our proposed method as well as numerical experiments demonstrating the ability of our method to sample the correct conditional measures in applications ranging from inverse problems to image in-painting.

Survival regression is used to estimate the relation between time-to-event and feature variables, and is important in application domains such as medicine, marketing, risk management and sales management. Nonlinear tree based machine learning algorithms as implemented in libraries such as XGBoost, scikit-learn, LightGBM, and CatBoost are often more accurate in practice than linear models. However, existing implementations of tree-based models have offered limited support for survival regression. In this work, we propose and implement loss functions for learning accelerated failure time (AFT) models in XGBoost, to increase the support for survival modeling for different kinds of label censoring. The AFT model assumes effects that directly accelerate or decelerate the survival time for different kinds of censored data sets. We demonstrate with real and simulated experiments the effectiveness of AFT in XGBoost with respect to a number of baselines, in two respects: generalization performance and training speed. Furthermore, we take advantage of the support for NVIDIA GPUs in XGBoost to achieve substantial speedup over multi-coreCPUs. To our knowledge, our work is the first implementation of AFT that utilizes the processing power of NVIDIA GPUs.

We introduce the novel concept of anti-transfer learning for neural networks. While standard transfer learning assumes that the representations learned in one task will be useful for another task, anti-transfer learning avoids learning representations that have been learned for a different task, which is not relevant and potentially misleading for the new task and should be ignored. Examples of such tasks are style vs content recognition or pitch vs timbre from audio. By penalizing similarity between the second network and the previously learned features, co-incidental correlations between the target and the unrelated task can be avoided, yielding more reliable representations and better performance on the target task. We implemented anti-transfer learning with different similarity metrics and aggregation functions. We evaluate the approach in the audio domain with different tasks and setups, using four datasets in total. The results show that anti-transfer learning consistently improves accuracy in all test cases, proving that it can push the network to learn more representative features for the task at hand.

Several sampling algorithms with variance reduction have been proposed for accelerating the training of Graph Convolution Networks (GCNs). However, due to the intractable computation of optimal sampling distribution, these sampling algorithms are suboptimal for GCNs and are not applicable to more general graph neural networks (GNNs) where the message aggregator contains learned weights rather than fixed weights, such as Graph Attention Networks (GAT). The fundamental reason is that the embeddings of the neighbors or learned weights involved in the optimal sampling distribution are changing during the training and not known a priori, but only partially observed when sampled, thus making the derivation of an optimal variance reduced samplers non-trivial. In this paper, we formulate the optimization of the sampling variance as an adversary bandit problem, where the rewards are related to the node embeddings and learned weights, and can vary constantly. Thus a good sampler needs to acquire variance information about more neighbors (exploration) while at the same time optimizing the immediate sampling variance (exploit). We theoretically show that our algorithm asymptotically approaches the optimal variance within a factor of 3. We show the efficiency and effectiveness of our approach on multiple datasets.